Musicnet preprocessing

src/musicNet/main.py

@@ -0,0 +1,27 @@
+import numpy as np
+import os
+import sys
+from sklearn.tree import DecisionTreeClassifier
+from sklearn.ensemble import RandomForestClassifier
+
+path = 'src/musicNet/processed_data'
+X_train = np.load(path + '/train_data_midi.npy')
+X_test = np.load(path + '/test_data_midi.npy')
+y_train = np.load(path + '/train_labels_midi.npy')
+y_test = np.load(path + '/test_labels_midi.npy')
+
+dt_clf = DecisionTreeClassifier(random_state=42)
+
+dt_clf.fit(X_train, y_train)
+training_accuracy = dt_clf.score(X_train, y_train)
+accuracy = dt_clf.score(X_test, y_test)
+print(training_accuracy)
+print(accuracy)
+
+rf_clf = RandomForestClassifier(random_state=42, max_features=512, n_estimators=100)
+
+rf_clf.fit(X_train, y_train)
+training_accuracy = rf_clf.score(X_train, y_train)
+accuracy = rf_clf.score(X_test, y_test)
+print(training_accuracy)
+print(accuracy)

src/musicNet/musicnet4641.py

@@ -38,11 +38,15 @@ def create_id_dict(df):
         id_dict[composers[i]] = df.loc[df['composer'].values == composers[i]]['id'].values
     return id_dict
 
-def trim_data_dict(data_dict, min_len=20):
+def trim_data_dict(data_dict, min_len=20, is_np_array=False):
     new_data_dict = {}
     for composer in list(data_dict.keys()):
-        if len(data_dict[composer]) >= min_len:
-            new_data_dict[composer] = data_dict[composer]
+        if is_np_array:
+            if data_dict[composer].shape[0] >= min_len:
+                new_data_dict[composer] = data_dict[composer]
+        else:
+            if len(data_dict[composer]) >= min_len:
+                new_data_dict[composer] = data_dict[composer]
     return new_data_dict
 
 def rand_id_sample(dict):